2

I am doing the following udemy course:

https://www.udemy.com/course/neural-network-understanding-and-building-an-ann-in-r/

and the instructor installs R version 3.5.7 saying 'download the latest version' but while following along, I found that version 4.0.2 is out and I am wondering if there are any compatibility issues between the 2 versions like python 3 and 2.

Which version should I download if I wanted to keep up with the course ?

and can I have multiple versions installed on my Linux machine and alternate between them ?

Also how much of a difference is there between the 2 versions?

Edit: Ok thanks everyone, ill install 4.0.2 and if I face any issues ill install 3.5.7 too. Also I found the course for free on a subreddit called freeudemycoupons if anyone is interested

https://www.reddit.com/r/FreeUdemyCoupons

whyisitnotworking
  • 41
  • 1
  • 5
  • 4
    Welcome to StackOverflow! You will be fine with version R 4.0.2. Difference are detailed in the NEWS file but you should not expect any (maybe with the exception of the random number generator seeding change in R 3.6.* if you want to reproduce draws done in R 3.5.*). Otherwise your ANN studies will work just the same -- go for it! – Dirk Eddelbuettel Jun 26 '20 at 19:38
  • You can have both versions on your computer too. It's pretty easy to switch to either version in RStudio – Tung Jun 26 '20 at 19:51
  • 1
    Another change in R-4.0 is a change from `data.frame(..., stringsAsFactors=FALSE)` to `=TRUE`. This *can* be a big factor if you don't know the difference and/or don't know how to work with `factor`s. They're not cosmically difficult, but new-to-programming users often stumble by errors, warnings, or unexpected results when a `factor` unknowingly exists. – r2evans Jun 26 '20 at 20:07

2 Answers2

4

This is very interesting question! I think that for online courses it's usually OK to use a slightly newer version of R. It is very unlikely (however, of course, not impossible) that an online course would use some feature that could be undergoing significant changes at that time. If you want to be sure about all new features and changed behavior in different R versions, you can read the official changelog.

Of course, one should be very careful when updating R in production, especially changing major version. It requires a lot of testing that everything runs correctly and you get proper results. But in online course you will probably get similar results and you can go on. And if you still get stuck, you can always ask here at Stack Overflow (or Stack Exchange: Data Science, AI) at no cost.

And for installing multiple versions of R on Linux you can use RStudio's instructions, which are pretty useful for various distributions and RStudio supports multiple versions and makes switching quite easy.

These threads could be useful:

1

We get output of different from R 4.0.2 and R 3.6.1 because of attach data for CSV file, for example, We get character and mode (Smoke, Gender, Caesarean) from 4.0.2 and yes, no and number from R 3.6.1.

Dr. Kyi Kyi Win
  • 11
  • 1